Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Videos

A multitarget training method for artificial neural network with application to computer-aided diagnosis.

Bei Liu¹, Yulei Jiang

¹Department of Radiation Oncology, University of Southern California, Los Angeles, CA, USA. beiliu@usc.edu

Medical Physics

|January 10, 2013

Summary

This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

Efficient separation of butane isomers via ZIF-8 slurry on laboratory- and pilot-scale.

Nature communications·2022

Same author

Influence of Sea Level Anomaly on Underwater Gravity Gradient Measurements.

Sensors (Basel, Switzerland)·2022

Same author

MW‑9, a chalcones derivative bearing heterocyclic moieties, attenuates experimental autoimmune encephalomyelitis via suppressing pathogenic T<sub>H</sub>17 cells.

Molecular medicine reports·2022

Same author

Evaluation of Emulsifying Ability of Phospholipids by Langmuir Monolayers and Stability of High Oil Ratio O/W Emulsions.

AAPS PharmSciTech·2022

Same author

Evaluation of Occupational Health and Safety Management of Listed Companies in China's Energy Industry Based on the Combined Weight-Cloud Model: From the Perspective of FPE Information Disclosure.

International journal of environmental research and public health·2022

Same author

Multisystem Langerhans Cell Histiocytosis in Younger Infants First Presenting in Skin: A Case Series.

Journal of personalized medicine·2022

A new multitarget training method for artificial neural networks (ANNs) improves breast cancer classification accuracy compared to traditional binary methods. This approach offers reduced output variability and better performance in computer-aided diagnosis.

Area of Science:

Artificial Intelligence
Medical Imaging
Machine Learning

Background:

Accurate classification of breast lesions as malignant or benign is crucial for effective cancer diagnosis.
Current artificial neural network (ANN) training methods often use simplified binary targets, potentially limiting diagnostic precision.

Purpose of the Study:

To introduce and evaluate a novel multitarget training method for ANNs in two-class classification tasks.
To assess the efficacy of this method in the context of breast lesion classification from mammograms.

Main Methods:

The study employed a multitarget training approach using detailed histological diagnoses as target values, reflecting posterior probabilities of malignancy.
Conventional binary training with simple malignant/benign targets was used as a comparison.

Related Experiment Videos

Monte Carlo simulations and a mammography study were conducted to evaluate the methods.

Main Results:

The multitarget training method demonstrated reduced variability in ANN outputs compared to the binary method.
Simulation studies indicated improved overall classification performance with the multitarget method, except in cases with extremely large training datasets.

Conclusions:

The proposed multitarget ANN training method shows promise for enhancing computer-aided diagnosis of breast cancer.
This technique offers a potential improvement over conventional binary training for classification tasks in medical imaging.